Quantifying potential confounders of panel-based tumor mutational burden (TMB) measurement. (April 2020)

• Record Type: Journal Article
• Authors: Budczies, Jan; Kazdal, Daniel; Allgäuer, Michael; Christopoulos, Petros; Rempel, Eugen; Pfarr, Nicole; Weichert, Wilko; Fröhling, Stefan; Thomas, Michael; Peters, Solange; Endris, Volker; Schirmacher, Peter; Stenzinger, Albrecht
• Is Part Of: Lung cancer. Volume 142(2020)
• Page Start: 114
• Page End: 119
• Publication Date: 2020-04
• Subjects: CDS coding sequence -- CRC colorectal carcinoma -- CV coefficient of variation -- FFPE formalin-fixed and paraffin-embedded -- FN false negatives -- FP false positives -- HNSCC head and neck squamous cell carcinoma -- ICI immune checkpoint inhibitor -- NSCLC non-small-cell lung cancer -- Mbp mega base pairs -- psTMB panel sequencing based TMB -- SNP single nucleotide polymorphism -- TMB tumor mutational burden; Tumor mutational burden -- TMB -- Panel sequencing -- Confounders -- Panel size -- Stochastic error; Lungs -- Cancer -- Periodicals; Lung Neoplasms -- Abstracts; Lung Neoplasms -- Periodicals; Poumons -- Cancer -- Périodiques; Lungs -- Cancer; Periodicals; Electronic journals; Electronic journals; 616.99424
• DOI: 10.1016/j.lungcan.2020.01.019
• Languages: English
• ISSNs: 0169-5002

Abstract

Highlights: Confounders of panel sequencing based tumor mutational burden (psTMB) were analyzed. Analysis of panel size, germline filtering, biological and technical variance. Stochastic error connected to panel size was the largest of all error contributors. Quality control, optimized laboratory workflows and bioinformatics are essential. Analysis framework is applicable to other complex biomarkers such as HRD. Abstract: Objectives: Retrospective data including subgroup analyses in clinical studies have sparked strong interest in developing tumor mutational burden (TMB) as a predictive biomarker for immune checkpoint blockade. While individual factors influencing panel sequencing based measurement of TMB (psTMB) have been discussed in the recent literature, an integrative study quantifying, comparing and combining all potential confounders is still missing. Material and methods: We separated different potential confounders of psTMB measurement including "panel size", "germline mutation filtering", "biological variance" and "technical variance" and developed a specific error model for each of these factors. Published experimental psTMB data were fitted to the error models to quantify the contribution of each of the confounders. The total psTMB variance was obtained as sum over the variance contributions of each of the confounders. Results: Using a typical large panel (size 1–1.5 Mbp) total errors of 57 %, 42 %, 34 % and 28 % were observed for tumors with psTMB of 5, 10, 20 and 40 muts/Mbp. Even for large panels, the stochastic error connected to the panel size represented the largest of all contributions to the total psTMB variance, especially for tumors with TMB up to 20 muts/Mbp. Other sources of psTMB variability could be kept under control, but rigorous quality control, best practice laboratory workflows and optimized bioinformatics pipelines are essential. Conclusion: A statistical framework for the analysis of complex, genomic biomarkers was developed and applied to the analysis of psTMB variability. The methods developed here can support the analysis of other quantitative biomarkers and their implementation in clinical practice.